Teenager invents inexpensive DR screening tool

Together with her brother and schoolmate, Kayva Kopparapu developed Eyeagnosis – a smartphone app trained on a convolutional neural network (CNN).{{quote-A:R-W:450-I:2-Q: The device is ideal for making screening much more efficient and available to a broader population, -WHO:Dr J Fielding Hejtmancik, NIH Ophthalmologist}}The team fed the CNN 34,000 retinal scans from the National Institute of Health (NIH) and programmed it to recognise the signs of DR.For Kopparapu it was more than just a project, but a personal desire to contribute to alleviating the plight of people losing sight due to DR.In particular, she was inspired by her grandfather in India who had recently been diagnosed with the condition.“The lack of diagnosis is the biggest challenge and in India, there are programs that send doctors into villages and slums, but there are a lot of patients and only so many ophthalmologists,” Kopparapu told IEEE Spectrum – the official magazine of the Institute of Electrical and Electronics Engineers.One challenge that Kopparapu faced was that many of the scanned images found in the database were of low or poor quality, however, with the use of the high-resolution camera and flash mechanism of the smartphone, the app was able to enhance the images.{{image3-a:l-w:360}}They overcame the obstacle through the use of a simple 3D printed lens, which concentrates light from the smartphone camera flash on the back of the eye.This challenge also proved to be a blessing ,given Kopparapu’s aim was to create an app that used a smartphone camera as the imaging tool.NIH ophthalmologist Dr J Fielding Hejtmancik said that while more clinical data needs to be collected before the syst can be used broadly, it also had great potential.“The device is ideal for making screening much more efficient and available to a broader population,” he said.So far Eyeagnosis has been trialled on five patients at Aditya Jyot Eye Hospital, in Mumbai, and in each case made an accurate diagnosis.